Coupled Caspase and N-End Rule Ligase Activities Allow Recognition and Degradation of Pluripotency Factor LIN-28 during Non-Apoptotic Development

Summary

Recent findings suggest that components of the classical cell death machinery also have important non-cell death (non-apoptotic) functions in flies, nematodes, and mammals. However, the mechanisms for non-canonical caspase substrate recognition and proteolysis, and direct roles for caspases in gene expression regulation remain largely unclear. Here we report that CED-3 caspase and the Arg/N-end rule pathway cooperate to inactivate the LIN-28 pluripotency factor in seam cells, a stem-like cell type in C. elegans, thereby ensuring proper temporal cell fate patterning. Importantly, the caspase and the E3 ligase execute this function in a non-additive manner. We show that CED-3 caspase and the E3 ubiquitin ligase UBR-1 form a complex that couples their in vivo activities, allowing for recognition and rapid degradation of LIN-28, and thus facilitating a switch in developmental programs. The inter-dependence of these proteolytic activities provides a paradigm for non-apoptotic caspase-mediated protein inactivation.

Blurb

The caspase CED-3 cleaves the LIN-28 protein to limit its activity during seam cell patterning in C. elegans. Weaver et al. show that this non-apoptotic function of CED-3 occurs in collaboration with a ubiquitin ligase of the Arg/N-end rule pathway, and that activities of both of these proteins are required for efficient recognition and degradation of LIN-28.

Introduction

Classic cell death machinery have important non-apoptotic functions including muscle development in mammals (Fernando et al., 2002), neuronal regeneration in C. elegans (Pinan-Lucarre et al., 2012), control of longevity in C. elegans (Yee et al., 2014), stress-responsiveness in C. elegans (Judy et al., 2013), and control of stemness in mammals and C. elegans (Fujita et al., 2008; Weaver et al., 2014). Non-apoptotic caspase activity, with as-yet unknown function, has also been observed using caspase-sensitive cellular reporters in Drosophila (Tang et al., 2015). Direct caspase targets for non-canonical functions are largely unknown. We reported that CED-3 caspase limits LIN-28 protein levels and was necessary for proper temporal cell fate patterning of seam cells, a stem cell-like cell type in C. elegans (Figure 1A) (Weaver et al., 2014). However, the molecular mechanisms underlying these non-apoptotic caspase functions remain largely unclear.

Figure 1. CED-3 Caspase Initiates Proteolytic Processing of LIN-28 but Requires Other Factors for LIN-28 Inactivation.

(A) Diagram of LIN-28 regulation by microRNAs and CED-3 caspase (Weaver et al., 2014). Other factors (Rougvie and Moss, 2013) omitted for simplicity.

(B) Examination of endogenous CED-3 caspase for site-specific cleavage of LIN-28::GFP fusion (Figures S1A and S1B, experimental setup). Proteolysis reactions from whole worm extracts taken from wild-type animals (+) or ced-3(−) mutants (−) with either a LIN-28::GFP transgene (Tg) or a LIN-28(D31A)::GFP point mutant transgene (D31A). Arrow, full-length LIN-28::GFP fusion protein; arrowhead, ~30 kDa processed product (Figure S1C, independent replicates).

(C) Quantitation of degradation experiments shown in (B) and Figure S1C. Mean values with standard deviations. *Significant, p values indicated below plot for degradation product at 6 hr, Unpaired, two-tailed t-test.

(D) Diagram of lin-28 gene with DxxD CED-3 cleavage motif. The Δ32M-lin-28 mutation removes residues 2–31 (including DxxD motif).

(E) Illustration of lin-28-dependent phenotypes (Rougvie and Moss, 2013).

(F) Seam cell counts at fourth larval stage of lin-28(−) animals without or with the Δ32M-lin-28 transgene or endogenous lin-28 Δ32M mutation by Cas9 (*Significant, p<0.0001, Mann-Whitney). Number of animals scored and median values indicated.

(G) Vulva phenotypes of lin-28(−) animals at adulthood without or with the Δ32M-lin-28 transgene or endogenous lin-28 Δ32M mutation by Cas9 (*Significant, p<0.0001, Fisher’s exact test). Number of animals scored indicated.

Caspases with both apoptotic and non-apoptotic functions must somehow differentiate cognate substrates to yield the necessary distinct cellular outcomes. Thus, dual function caspases must be able to dynamically alter client substrate interactions through mechanisms that are currently unknown. Based on structural features of the LIN-28 caspase cleavage site that we previously identified in vitro, we suspected that other proteolytic systems may be involved in LIN-28 inactivation (Weaver et al., 2014).

The N-end rule system has been elegantly decoded into distinct structural features and enzymatic processes, including the N-terminal degradation signals (N-degrons) and the machinery needed to generate and then recognize the N-degron (Bachmair et al., 1986; Xia et al., 2008; Varshavsky, 2011).

We report here a coupled function between a caspase and the Arg/N-end rule E3 ubiquitin ligase pathway in a gene expression mechanism that controls temporal cell fate patterning of seam cells. We provide evidence that the CED-3 caspase and UBR-1 form a complex that serves to rapidly inactivate LIN-28 protein in vivo to support robustness and thus dynamically shift stage-specific developmental programs. Importantly, the requirement of the UBR-type E3 ligase by CED-3 caspase to efficiently initiate LIN-28 degradation points to a paradigm for differential recognition of a non-apoptotic target protein.

Results

CED-3 Cleaves LIN-28 to Accelerate Its Degradation

We showed purified CED-3 caspase can partially cleave LIN-28 at a canonical DxxD cleavage motif DVVD (residues 28–31) in vitro and that loss of CED-3 caspase function leads to delayed down-regulation of LIN-28 during larval development in C. elegans (Weaver et al., 2014). Here, we wanted to test the necessity for this cleavage in initiating LIN-28 degradation.

CED-3-mediated proteolysis of a LIN-28::GFP fusion protein was assayed using whole animal extracts (Figures S1A and S1B). Based on physical and chemical studies, the GFP fold is relatively stable (Huang et al., 2007). We found that endogenous CED-3 caspase enhances proteolytic processing of LIN-28 at Asp31, based on accumulation of a C-terminal GFP moiety (Figures 1B, and S1C). Accumulation of the degradation product is significantly delayed at 6 hours by ced-3(−) and lin-28(D31A) mutations while other proteolytic processes continued to act (diminishment of the actin control with time) (Figures 1C and S1C). The processed form (~30 kDa) is significantly smaller than LIN-28::GFP lacking the first 31 residues (~46 kDa), indicating that most of the LIN-28 protein was degraded beyond the CED-3 cleavage at Asp31, leaving the more stable GFP moiety. These results indicate endogenous CED-3 caspase plays a prominent role in initiating LIN-28 degradation.

CED-3 Requires Other Proteolytic Factors to Inactivate LIN-28

We used two approaches to test if the truncation produced by CED-3 cleavage directly inactivates LIN-28 function. First, we made a synthetic transgene where amino-acids 2 through 32 were deleted (Figure 1D, Δ32M-lin-28). Second, we made this same deletion in the native lin-28 locus using CRISPR-Cas9 mutagenesis (Paix et al., 2014; Paix et al., 2015). We then evaluated lin-28 function based on lin-28-dependent phenotypes (Moss et al., 1997; Ambros, 2000; Rougvie and Moss, 2013) (Figure 1E). We found that both seam cell number and vulval development were restored to nearly wild-type levels by the transgene [in a lin-28(−) background] or the engineered endogenous mutation (Figure 1F and 1G), indicating that the Δ32M-LIN-28 mutation was functional. These results suggested that CED-3 must collaborate with another proteolytic mechanism to inactivate LIN-28.

CED-3 Cleavage of LIN-28 Generates an Arg/N-End Rule N-degron

CED-3 is an endopeptidase, making the Arg/N-end rule pathway a prominent candidate collaborator that promotes LIN-28 degradation, as it determines protein stability largely as a function of the N-terminal residue (Varshavsky, 2011). As depicted in Figure 2A, substrates of the Arg/N-end rule system have three critical determinants (Bachmair et al., 1986; Bachmair and Varshavsky, 1989; Chau et al., 1989; Prakash et al., 2004; Varshavsky, 2011).

Figure 2. CED-3 Cleaved LIN-28 Is An Arg/N-End Rule Substrate and UBR-1 Ligase Works Non-Additively with CED-3 Caspase to Regulate LIN-28-Dependent Seam Cell Divisions.

(A) Illustration of Arg/N-end rule substrates.

(B) Diagram of ubiquitin fusion technique. X indicates the N-terminal residue following cleavage by deubiquitinase.

(C) Phosphorscreen image of ³⁵S-labeled proteins following in vitro N-end rule assays.

(D) Log scale quantitation of three independent in vitro N-end rule assays (Figure S1D, independent replicates and Figure S1E using CED-3 cleaved LIN-28 directly as input). Means, dots. Standard deviations, brackets

(E–F) Analysis of ubr-1(−) on seam cell divisions in L4 stage animals (Figures S2A–S2E, ubr-1 homology and mutation). Pseudo-colored images in (E) show seam cell nuclei. Scale bars in (E), 50 μm. Total numbers of animals scored indicated. Orange bars, median values (given below each dot plot). Hatched line, 16 seam cells typically found in wild-type animals (same throughout). **Significant compared to single mutants, p<0.001, Mann-Whitney test. (Figures S3A–S3C, additional data on seam cell specification).

(G) Analysis of ate-1(RNAi) on seam cell divisions in L4 stage animals. *Significant, p <0.0001, ain-1(−) (mock RNAi) compared to wt (mock RNAi); **Significant, p <0.0001, ain-1(−); ate-1(RNAi) compared to ain-1(−) (mock RNAi) and wt treated with ate-1(RNAi)(Mann-Whitney). Number of animals scored indicated.

(H) Analysis of ubr-1(−) and ced-3(−) to determine impact on seam cell number [**Significant compared to single mutants, p<0.0001, ‡Also significant compared to ubr-1(−);ced-3(−), p = 0.0003, but not significant compared to ain-1(−);ubr-1(−) or ain-1(−);ced-3(−), Mann-Whitney test]. Data independent of panel (F) (Figure S3D, genetic models).

(I) Analysis of ubr-1(−) and ced-3(−) to determine impact on adult alae formation. Percentages of gapped and low quality alae indicated [**Significant compared to single mutants, p<0.05, ‡Also significant compared to ubr-1(−);ced-3(−) but not significant compared to ain-1(−);ubr-1(−), Chi-square test].

To test whether Asn-LIN-28 meets all three structural requirements and is indeed a substrate for the Arg/N-end rule pathway, we used the ubiquitin fusion method (Brower et al., 2013). Asn32-LIN-28 (or the Met32-LIN-28 control) was made in frame with an N-terminal ubiquitin fusion (Figure 2B) and used in the in vitro N-end rule assay (Brower et al., 2013). We found that Asn-LIN-28 was quickly degraded by the Arg/N-end rule pathway (Figures 2C and S1D). By one hour, fifty percent of the initial Asn-LIN-28 was eliminated relative to the DHFR-Ub moiety internal control (log scale in Figure 2D). By comparison, the Met-LIN-28 negative control was considerably more stable (Figures 2C, 2D, S1D). Additionally, the modest Asn-LIN-28 product generated directly by in vitro CED-3 cleavage, is also destabilized by the in vitro N-end rule assay (Figure S1E). These results support a possible role for the Arg/N-end rule pathway in destabilizing Asn-LIN-28 in vivo.

C. elegans UBR-1 Is Homologous to Canonical UBR-type E3 Ubiquitin Ligases

The UBR box is a critical determinant for substrate recognition in the Arg/N-end rule pathway (Matta-Camacho et al., 2010). Alignments of UBRs from distinct species show that C. elegans UBR-1 has a high degree of homology with yeast UBR1 and human UBRs 1 and 2 (Figure S2A). Key residues within the given functional domains are highly conserved (Figure S2A). Sequence analysis showed that C. elegans UBR-1 has absolute conservation of both the atypical zinc finger (CX₂₄CX₂CX₂₁CXCX₁₁CX₂H) and the second zinc finger (CX₂CX₂₀HX₂H) in the UBR box as well as complete conservation of the negatively charged pocket and high conservation of the penultimate pocket within the UBR box (Figure S2A). Two residues in the penultimate pocket of C. elegans UBR-1 differ from human UBRs 1 and 2 (DGTCVM instead of DPTCVL). However, sequence examination shows that these positions are highly permissive across other human UBRs. Based on the high degree of homology with canonical UBRs, the C. elegans UBR-1 protein likely has analogous function in the Arg/N-end rule pathway.

UBR-1 Functions in Seam Cell Patterning

We generated a ubr-1(tm5996) allele that should severely disrupt UBR-1 function by deleting the DNA sequences encoding critical determinants of the N-recognin motif including half of the UBR box and nearly the entire ClpS core domain (Figure S2B). We cloned the ubr-1 cDNA from wild-type and ubr-1(tm5996) mutants and confirmed that the third exon, encoding critical UBR motifs, was skipped (Figures S2B–D). Using the crystal structure from the human UBR1 UBR box (Matta-Camacho et al., 2010) (PDB file 3NY1), we annotated the disruptions caused by the tm5996 mutation to the C. elegans UBR-1 protein sequence including loss of residues within the β2 strand, loop 2, the negatively charged pocket and aromatic amino-acid cluster (blue strands and spheres in Figure S2E). Loss of these residues in the substrate recognition pocket should thus represent a strong loss-of-function allele that we tested for effects on seam cell patterning.

With each larval molt in C. elegans, the lateral seam cells (V1–V4 and V6) divide asymmetrically with additional stem cells produced in the L2 stage by one symmetric division that duplicates only the V1–V4 and V6 seam cells (Sulston and Horvitz, 1977; Ambros and Horvitz, 1984). By adulthood, wild-type animals complete seam cell divisions and fully differentiate the resulting 16 seam cells on each side of the body (Joshi et al., 2010). Normal seam cell division and differentiation processes result in continuous adult alae. Following the L2-specific symmetric seam cell division, LIN-28 is rapidly down-regulated and a dynamic shift in developmental programs results in subsequent asymmetric seam cell divisions at later molts (Moss et al., 1997).

Loss of ain-1/GW182 function alone results in a mild increase from 16 to around 18–19 seam cells in L4 stage animals (Ding et al., 2005; Weaver et al., 2014). AIN-1 is a critical factor of the microRNA-induced silencing complex (miRISC) and has a well-established role in negatively regulating the expression of lin-28 related to seam cell specification. This has been demonstrated in multiple ways, including suppression of precocious timing defects of lin-28(−) by ain-1(−) (Ding et al., 2005), suppression of the ain-1(−);let-7 family(−) supernumerary seam cell phenotype by lin-28(−) mutation (Vadla et al., 2012), and increased lin-28 mRNA levels with compromised miRISC (Bagga et al., 2005). Thus, the ain-1(−) mutation provides a convenient background to test if other genes may function to limit lin-28 expression by observing whether there is a synergistic increase in seam cell numbers (Morita and Han, 2006; Weaver et al., 2014). We found that loss of ced-3 function enhanced ain-1(−) mutants for excessive seam cell proliferation (Weaver et al., 2014). Thus, if the Arg/N-end rule is necessary for eliminating LIN-28 function, compromising this pathway should also increase seam cell proliferation.

We found that the ubr-1(tm5996) allele synergistically enhanced the excessive seam cell proliferation phenotype of ain-1(−) (Figures 2E and 2F). Similar enhancement was observed by ubr-1(RNAi) treatment of ain-1(−) mutants (Figure S3A), which confirmed that ubr-1(tm5996) is a loss-of-function allele [hereafter called ubr-1(−)]. Additionally, ubr-1(−) was observed to enhance the ain-1(−) gapped adult alae phenotype that also marks a delay in the temporal cell fate patterning of seam cells during larval developmental progression (Zhang et al., 2007; Ding et al., 2005) (p <0.05, Chi-square test) (Figures S3B and S3C).

We also tested the role of the ate-1 gene, which encodes the C. elegans homolog of ATE1 arginyltransferase known to play a critical role in the Arg/N-end rule pathway (Varshavsky, 2011), in regulating seam cell divisions. We found that ate-1(RNAi) enhanced the ain-1(−) mutant defect leading to a significant increase in supernumerary seam cells (p <0.001, Mann-Whitney) (Figure 2G). These results suggested a role for the Arg/N-end rule pathway in preventing supernumerary L2-like symmetric divisions in later larval stages. These findings are comparable to ced-3(−) mutants we previously reported (Weaver et al., 2014).

UBR-1 and CED-3 Regulate Seam Cell Divisions in a Non-Additive Manner

We then analyzed the functional relationship between CED-3 and UBR-1 in regulating seam cell divisions. If CED-3 and UBR-1 limit LIN-28 through mechanisms that are independent of each other, we would expect the ced-3(−);ubr-1(−) double mutant to display a synergistic outcome resulting in excessive seam cell proliferation and subsequent gapped adult alae similar to the defects in the ain-1(−);ubr-1(−) or the ain-1(−);ced-3(−) double mutants (Figure S3D). However, we found that loss of both ubr-1 and ced-3 does not show synergism in the accumulation of L4 seam cells or gapped adult alae (Figures 2H and 2I). To exclude the possibility that the lack of enhancement was due to the limitation by a potential phenotype threshold, we also built a ubr-1(−);ced-3(−);ain-1(−) triple mutant to examine the potential additive effect between ubr-1(−) and ced-3(−) in the ain-1(−) background that may present a more sensitized assay condition. The results for the triple mutant compared to the two double mutants confirmed the lack of an additive effect between ubr-1(−) and ced-3(−) (Figures 2H and 2I). Since loss of either ced-3 or ubr-1 function is the same as loss of both in generating the seam-related phenotypes, these data imply strong inter-relatedness of the caspase and UBR-type ligase functions.

UBR-1 Negatively Regulates LIN-28 Expression In Vivo

Using an antibody that recognizes the C-terminus of endogenous LIN-28 (Weaver et al., 2014), we saw that loss of ubr-1 function resulted in an obvious delay in the down-regulation of LIN-28 during the period of 20 to 30 hours post-feeding, as animals transitioned through mid-larval development (Figures 3A and S3E). The ubr-1(−) mutant showed a 3-fold increase in whole-animal LIN-28 protein levels at around 30 hours (Figure 3B, left panel). Upregulation of LIN-28 could be observed if UBR-1 regulated lin-28 expression by impacting lin-28 transcription or RNA stability. To address this possibility, we examined lin-28 mRNA levels and found no significant increase in ubr-1(−) mutants (Figure 3B, right panel). The most straightforward explanation for these data is that the UBR-1-mediated Arg/N-end rule pathway is involved directly in LIN-28 degradation in late larval stages. Surprisingly, full-length LIN-28, and not the expected Asn-LIN-28 CED-3 cleavage product, accumulated in the ubr-1(−) mutants, suggesting that UBR-1 is necessary in vivo for CED-3 cleavage at Asp31.

Figure 3. Inter-Dependent UBR-1 and CED-3 Activities Limit LIN-28 Expression.

(A) Western blot of endogenous LIN-28. Arrows, A and B isoforms.

(B) Left panel: quantitation of full-length LIN-28 at 30h (Figure S3E, independent replicates) with mean and standard deviation (wt mean set to 1.0) (*Significant, p = 0.0476, unpaired t test with Welch’s correction). Right panel: quantitation of lin-28 mRNA levels by qRT-PCR (p=0.4899, unpaired t test with Welch’s correction).

(C–E) Analysis of LIN-28::GFP fusion protein expression during third larval stage. Box-whisker plot in (C) of gonad lengths to confirm stage-matching of input animals (n = 20 for each genotype). Images in (D) are differential interference contrast (DIC) and pseudo-colored LIN-28::GFP. Scale bars in (D), 50 μm. Scatter plot in (E) shows LIN-28::GFP intensity as a function of gonad length [data from (C)].

(F) Comparisons of p values for results given in (E). *Significant, p <0.05, wt compared to either single mutant or double mutant, Mann-Whitney test.

(G) Proteolysis reactions of whole worm extracts taken from either ubr-1(−) mutant (−) or ubr-1 wild-type (+) animals with an integrated LIN-28::GFP transgene (Tg) (Figure S3F, independent replicates). Arrow in (G), full-length LIN-28::GFP fusion protein; arrowhead, ~30 kDa processed product.

(H) Quantitation of three independent degradation experiments shown in (G) and Figure S3F. Mean values with standard deviations shown (*Significant, p values indicated, wt compared to ubr-1(−) at the indicated times, Unpaired, two-tailed t-test).

UBR-1 and CED-3 Activities Are Coupled in LIN-28 Repression

Since loss of both ubr-1 and ced-3 did not show synergism in seam-related defects (Figures 2G and 2H), we wanted to test if the ubr-1(−);ced-3(−) double mutant had synergistic accumulation of LIN-28. We analyzed individual animals matched at the mid-larval stage for LIN-28::GFP expression by DIC optics (Figures 3C–3E). We used gonad length to obtain a similar distribution of mid-larval stage animals as gonad extension is not grossly affected by lin-28 activity (Ambros and Horvitz, 1984; Moss et al., 1997; Abbott et al., 2005). Animals were then evaluated for hypodermal LIN-28::GFP expression (Figure 3D). Calculating LIN-28::GFP hypodermal expression intensity as a function of gonad length confirmed that LIN-28::GFP expression was significantly upregulated in ubr-1(−) mutants, ced-3(−) mutants, and the ubr-1(−);ced-3(−) double mutants relative to wild-type (p <0.0044, Mann-Whitney) (Figures 3E and 3F). However, the single and double mutants did not show significant differences from each other (p >0.094, Mann-Whitney) (Figures 3E and 3F).

Since ubr-1(−) mutation alone showed upregulation of full-length LIN-28 without obvious accumulation of CED-3 processed LIN-28 (Figures 3A and S3E) and since the ubr-1(−);ced-3(−) double mutant did not show synergistic accumulation of LIN-28 (Figure 3E), these findings suggest a model where the biochemical activities of CED-3 and UBR-1 are inter-dependent and coupled. To test this possibility, we used the whole worm extract proteolysis assay to assess whether loss of ubr-1 function alone could act analogously to ced-3(−) and delay the accumulation of the degradation intermediate shown in Figures 1B and S1C. We found that ubr-1(−) had a significant delay in the accumulation of the GFP degradation intermediate (Figures 3G, 3H and S3F).

These LIN-28 protein expression and degradation data suggest that UBR-1 and CED-3 may require each other to repress LIN-28 levels, which is supported by the genetic interaction data presented earlier (Figures 2G and 2H). In other words, UBR-1 and CED-3 likely execute their roles in the same functional complex.

UBR-1 and Arginyltransferase ATE-1 Complex with CED-3 Caspase

To determine if UBR-1, ATE-1 and CED-3 form a complex in cells, we expressed the C. elegans coding sequences for CED-3, ATE-1 and UBR-1 in HEK293 cells (Figures 4A–4E). Since expression of full-length CED-3 is lethal in HEK cells (Figure S4A), we used the C358S active site point mutation (Xue et al., 1996). Using co-immunoprecipitation and Western blot analysis, we found that full-length CED-3 was able to pull down full-length UBR-1 (Figures 4A–4B). Since CED-3 auto-processes itself into various domains and subunits (Xue et al., 1996) (Figure 4C), we decided to subclone these fragments individually to test their ability to specifically associate with UBR-1. We found that the p17 sequence was required for UBR-1 interaction (p17 is contained in the full-length protein and the p32 domain) (Figures 4D, S4B, S4C, and S4D). These findings support the formation of a CED-3-UBR-1 complex via the p17 subunit. This observation is even more compelling since the CED-3 p17 subunit was expressed at a lower level than the p15 subunit and the Pro-domain that do not physically interact with UBR-1 (Figures 4D, S4C, and S4D). Moreover, we tested the ability of CED-3 to interact with two isoforms of ATE-1 (that differ by an alternatively spliced third exon) and found that the canonical ATE-1A isoform strongly interacted with CED-3 regardless of whether UBR-1 was present or absent (Figure 4E). This finding indicated that CED-3 interacted with N-end rule components in a specific manner.

Figure 4. Caspases Likely Form Complexes with Arg/N-end rule Components.

(A–B) Co-immunoprecipitation (Co-IP) analyses for UBR-1 and CED-3 caspase. The CED-3(C358S) mutation prevents cell death (Figure S4A). Tag sequences: see STAR Methods. WB: Western blot. Two tags were used with CED-3 to demonstrate the specificity of this interaction since UBR-1 tags are sterically inaccessible for immunoprecipitation (explanation in STAR Methods).

(C) Diagram of CED-3 auto-processing (Xue et al., 1996) indicated (black arrowheads). The p15 domain can be further processed to a p13 subunit. Red asterisk marks Cys residue in active site.

(D) Co-IP analyses with Western blot (WB) showing that the p17 subunit specifically associates with UBR-1. The auto-processed domains and subunits were cloned individually and tested for interaction with UBR-1 (Figures S4B–S4D, independent replicates).

(E) Co-IP analyses to show the interaction between full-length ATE-1 and CED-3 caspase.

(F) Co-IP analyses of human caspases and human UBR2 (Figure S4E, independent replicate and Figure S4F, caspase homology).

(G) Model for a complex containing CED-3 caspase and Arg/N-end rule components.

Human UBR2 and Human Caspase 8 Form A Complex

To assess the possible conservation of caspase/UBR interactions, we cloned multiple human caspases into a mammalian expression vector and made active site Cysteine to Serine mutations. Caspase 1 is an inflammatory caspase (Riedl and Shi, 2004). Caspases 3, 6, and 7 are executioner caspases (Riedl and Shi, 2004). Caspases 2 and 8 are initiator caspases (Riedl and Shi, 2004). We also cloned full-length human UBR2 (Varshavsky, 2011) (Figure S2A). We found that Caspase 8 repeatedly had the strongest interaction with UBR2 (Figures 4F and S4E). It should be noted that we routinely had difficulty expressing human Caspase 2 (Figures 4F and S4E). Our data suggest the possibility of functionally important caspase/UBR interactions across metazoans.

Discussion

By uncovering coupled activities of CED-3 caspase and an N-end rule protein degradation system in limiting the level of LIN-28 for accurate stem cell behavior and developmental timing in C. elegans, this study provides a mechanistic paradigm for the under-investigated, non-canonical roles of apoptotic caspases in regulating gene expression for robustness of non-apoptotic physiological functions. Caspases must cleave different sets of targets to achieve mutually exclusive apoptotic or non-apoptotic functions, and our findings suggest that forming a complex with Arg/N-end rule components may critically alter target recognition and processing (Figure 4G). It is also conceivable that such functions may serve to activate gene expression as caspase-mediated proteolysis has precedent for target activation.

The finding that human Caspase 8 interacted strongly with human UBR2 is intriguing since human Caspase 8 has been implicated in non-apoptotic functions of monocyte differentiation and trophoblast fusion (Black et al., 2004; Sordet et al., 2002). Human Caspase 8 is similar to C elegans CED-3 in domain architecture (Figure S4F). The broad functional significance of conserved UBR-type E3 ligase-caspase interactions remains an exciting area of future study.

Moreover, we found that CED-3 also physically interacts with ATE-1 (with or without UBR-1), supporting a case for a functional complex containing at least these three proteins (Figure 4G). We do not know whether these three proteins represent the sufficient components to inactivate LIN-28. Further study should address the composition of the complex and potential interactions between ATE-1, UBR-1, and other Arg/N-end rule components within the complex and their relationship with CED-3. The physical interaction data, together with the expression data that showed the accumulation of full-length LIN-28 in a ubr-1(−) mutant (Figures 3A and S3E) and a lack of synergistic or additive effects between mutations in ced-3 and ubr-1 genes (Figures 2H and 2I), suggest a model where the functions of CED-3 and UBR-1 are coupled and inter-dependent for the efficient recognition and destruction of LIN-28 (Figure 4G).

We propose that this complex allows for a net physiological outcome of enhanced specificity and efficiency of LIN-28 destabilization at a critical developmental time when seam cells are transitioning out of a proliferative state. Additional structural studies may be required to fully understand how the protein-protein interactions of CED-3 with N-end rule components affect cognate substrate recognition and processing. It should be noted that loss of ubr-1 does not fully eliminate accumulation of the LIN-28 degradation product (Figure 3G) similar to loss of ced-3 (Figure S1C) implying potential other proteolytic processes in a homogenized lysate.

Previous studies showed that Drosophila and mammalian apoptotic caspase functions can be inhibited by the Arg/N-end rule pathway to prevent spurious cell death (Ditzel et al., 2003; Piatkov et al., 2012). Moreover, caspase-cleaved peptides (pro-apoptotic effectors) were shown to accumulate when the Arg/N-end rule was blocked (Piatkov et al., 2012). In contrast to this antagonistic relationship between caspases and the Arg/N-end rule, we find that a non-apoptotic caspase function is coupled to the Arg/N-end rule in the degradation (rather than accumulation) of LIN-28, as we show UBR-1 was required in vivo for efficient site-specific caspase cleavage. Therefore, we do not expect to observe deamidated or arginylated LIN-28 intermediates in the absence of ATE-1 or UBR-1. Our findings, in light of the previous studies, are highly intriguing and suggest that very dynamic interactions between caspases and the N-end rule under different physiological conditions in particular cells or tissues can result in distinct outcomes.

LIN-28 is a critical determinant of cell fate, with regulatory inputs at multiple levels. The combined functions of microRNAs/miRISC acting at the mRNA level (Moss et al., 1997; Zhang et al., 2007; Zinovyeva et al., 2014) and the CED-3/UBR-1 complex acting at the protein level allow for rapid and dynamic alteration of gene expression for proper execution of developmental programs in a stem-like cell type. It is interesting that individual components of the miRISC and the caspase/UBR proteolytic system are not strictly essential under common laboratory culturing conditions, but together support robust development with very accurate and precise developmental timing transitions. Such semi-redundancy in regulation may be important to buffer rapid alterations in environmental conditions or nutrient availability during critical developmental events.

STAR METHODS

CONTACT FOR REAGENT AND RESOURCE SHARING

Further information and requests for reagents may be directed to, and will be fulfilled by the corresponding author Benjamin Weaver (benjaminpweaver@gmail.com).

EXPERIMENTAL MODEL AND SUBJECT DETAILS

C. elegans strains used in this study

The wild-type (wt) reference strain is N2 Bristol. The ced-3(−) mutant bears the ced-3(n717) allele and is the reference allele for ced-3 loss-of-function (Ellis and Horvitz, 1986). The lin-28::gfp and lin-28(D31A):gfp integration strains were previously reported (Moss et al., 1997; Weaver et al., 2014). The lin-28(−) mutant bears the lin-28(n719) allele (Moss et al., 1997). The Δ32M-lin-28 transgene (detailed below) was used in lin-28(n719) mutants to test restoration of lin-28 function. We also generated the Δ32M-lin-28 mutation in the endogenous lin-28 locus using a CRISPR-Cas9 method (Paix et al., 2014; Paix et al., 2015) (see below). The JR667 strain with seam cell reporter was used to mark seam cells (Joshi et al., 2010). The ain-1(−) mutant bears the ain-1(ku322) allele previously reported (Ding et al., 2005). The ubr-1(−) mutant bears the ubr-1(tm5996) allele that was generated by a random mutagenesis method with TMP/UV. The mutation was confirmed by sequence analysis using primers P1 and P2 (see Table S1 for all oligos). The allele is an indel with 457 bp deletion and a 17 bp insertion in the genomic DNA. This mutation eliminates most of Exon 3 that encodes critical portions of the UBR-type zinc-finger and ClpS core domain and leads to exon skipping (Figures S2B–S2E). The phenotypes observed for ubr-1(tm5996) were similar to ubr-1 RNAi (Figure S3A) confirming that ubr-1(tm5996) is a loss-of-function allele.

Cell culture

HEK 293 cells were used only as a protein expression system. Cells were obtained from ATCC and were maintained in a humidified cabinet at 37°C with 5% CO₂. Cells were cultured with DMEM supplemented with 10%FBS, 4 mM L-Glutamine, 100 Units penicillin per mL, 100 μg streptomycin per mL, and 0.25 μg amphotericin B per mL.

METHOD DETAILS

Plasmids, cDNA cloning, constructs, epitope tagging, oligos and qPCR primers

All primers and oligos are detailed in Table S1. The LIN-28::GFP integration was previously detailed (Moss et al., 1997). The LIN-28(D31A)::GFP integration was also previously detailed (Weaver et al., 2014). Briefly, the LIN-28(D31A)::GFP transgene was made using the LIN-28::GFP transgene as the parental DNA with Q5 point mutagenesis (NEB) such that the LIN-28A Asp31Ala mutation is the only difference in the constructs. The Δ32M-LIN-28 was generated from the wild-type lin-28::gfp gene cloned into pBSIIKS(−) with primers P3 and P4 with Q5 mutagenesis to delete the first 31 amino-acids and replace the N32 with a Met residue.

Plasmid pCB268 (Brower et al., 2013) was a gift from Alexander Varshavsky (Addgene plasmid 53363). Plasmid c-Flag pcDNA3 was a gift from Stephen Smale (Addgene plasmid 20011). For the Ub fusion technique (Figures 2B–2D and S1D–S1E), the nFlag::DHFR::Ub::SacII sequence was moved from pCB268 to pTNT (Promega) using primers P5 and P6. Asn32-LIN-28 and Met32-LIN-28 control were fused in frame with the N-terminal ubiquitin moiety in the pTNT_nFlag::DHFR::Ub::SacII vector using primers P7, P8, and P9 generating pWE517 (pTNT_nFlag::DHFR::Ub::Asn 32 LIN-28::cFlag) and pWE518 (pTNT_nFlag::DHFR::Ub::Met 32 LIN-28::cFlag), respectively. Primers P10 and P11 were used to amplify the LIN-28A coding sequence from C. elegans cDNA into pTNT using MluI and KpnI sites (this construct was only used in CED-3 cleavage Figure S1E).

To make pcDNA3_UBR-1::V5, we replaced both the multiple cloning site (MCS) and the c-terminal Flag tag in the pcDNA3_c-Flag vector with another MCS and V5 tag (GKPIPNPLLGLDST) by annealing complimentary oligos P12 and P13 then ligating into the HindIII and EcoRI sites of the original vector generating a HindIII_NheI_AfeI_KpnI_KasI_V5 Tag_Stop_EcoRI sequence. The 6.2 kb ubr-1 coding sequence was amplified with primers P14 and P15 from C. elegans cDNA (P14 added a Kozak consensus sequence to assist translation of the C. elegans protein). The amplimer was ligated into the new MCS with KpnI and KasI sites.

Full-length CED-3 was amplified from C. elegans cDNA using primers P16 and P17 and inserted into pcDNA3_c-Flag (Addgene, 20011) to add a c-terminal Flag tag (DYKDDDDK) (primer P16 added a Kozak sequence). The C358S active site mutation was made by Q5 mutagenesis using primers P18 and P19. The pro-domain with Flag tag was made by Q5 mutagenesis using primers P20 and P21. The p32 domain with Flag tag was made by Q5 mutagenesis using primers P22 and P23. The p17 domain with Flag tag was made by Q5 mutagenesis using primers P24 and P25. The p15 domain with Flag tag was made by Q5 mutagenesis using primers P26 and P27. The C-terminal Flag tags in full-length CED-3, the p15 domain and the p32 domain were replaced with HA tags (YPYDVPDYA) by Q5 mutagenesis using primers P28 and P29. The C-terminal Flag tag in p17 was replaced with an HA tag by Q5 mutagenesis using primers P30 and P31. The C-terminal Flag tag in the pro-domain was replaced with an HA tag by Q5 mutagenesis using primers P32 and P33.

Full-length ATE-1A and D isoforms were amplified from C. elegans wild-type cDNA using primers P34 and P35 and inserted into pcDNA3_c-Flag using HindIII and KpnI sites.

Full-length coding sequences of human Caspase 1, Caspase 3 and Caspase 6 were subcloned from plasmids pET21b-Casp1-His (Addgene #11809), pET23b-Casp6-His (Addgene#11823) and pcDNA3-Casp3-myc (Addgene #11813), respectively, using primers P36–P41 into pcDNA3_c-HA. C-terminal HA tags replaced C-terminal Flag tags for Caspase 2 and Caspase 7 by modifying pcDNA3-Caspase-2-Flag (Addgene #11811) and pcDNA3-Caspase-7-Flag (Addgene #11815) with primers P42–P45. A C-terminal HA tag was added to pcDNA3-Caspase-8 (Addgene #11817) using Q5 mutagenesis with primers P46–P47. The active site Cysteines were then converted to Serines using Q5 mutagenesis with primers P48–P60.

The full-length coding sequence of human UBR2 was cloned by assembly of 3 fragments from cDNA of NCI H716 cells using primers P60–P66. NEBuilder® HiFi DNA Assembly Master Mix (NEB) was used to assemble 3 fragments into pCDNA3_cV5.

The Δ32M-lin-28 mutation in the endogenous lin-28 locus was generated by modification of a CRISPR-Cas9 method (Paix et al., 2014; Paix et al., 2015). The oligos used in this construction were P67–P70 to make the sgRNA expressing plasmid (Dickinson et al., 2013). Large ssDNAs used for CRISPR-Cas9 were U1 for Δ32M-lin-28 repair and U2 for dpy-10(cn64) co-conversion.

qPCR primers to measure rpl-4 mRNA levels were P71–P72. qPCR primers to measure lin-28 mRNA levels were P73–P74.

L1 stage, whole worm extract proteolysis reactions

The whole worm extract proteolysis assay used for Figures 1B, 3G, S1C, and S3F was done as follows. Large preparations of eggs were obtained from the indicated strains by alkaline hypochlorite dissolution of gravid adults, washed in M9 solution, and allowed to hatch at 20°C. Synchronous L1-stage animals were collected by centrifugation, flash-frozen, and stored at −80°C. Frozen pellets were homogenized by sonication in a small volume of lysis buffer (10 mM Tris, pH 7.4, 1 mM EDTA, and 0.5% NP-40, in the absence of protease inhibitors). Supernatants from centrifugation were then measured for total protein by BCA assay. Degradation assays were assembled with the same amount of total protein input for each sample in a given assay (at 0.4 mg/ml final concentration for each experiment) and adjusted to 150 mM NaCl. This protein concentration was chosen to prevent over-proteolysis. Reactions were incubated at 25°C and volume-equivalent samples were collected at 0, 3, and 6 hours (as indicated) from the same initial reaction tube. Collected time points were stopped from further degradation by addition of SDS sample buffer. Samples (same amount of input equivalent protein per well) were resolved on a denaturing SDS-PAGE gradient gel and visualized by Western analysis. Quantitation of bands was performed with Fiji (Image J) such that the intensity of bands within a lane was summed to 1 and the fraction of the degradation band was expressed as the percent of total density for the given sample at a given time.

Western blot, antibodies, and co-Immunoprecipitation

Biological replicates were all independently generated, collected and processed. The anti-GFP antibody (Figures 1B, 3G, S1C, and S3F) was commercially obtained (Clontech, JL-8 mouse monoclonal). The anti-actin antibody (Figures 1B, 3A, 3G, S1C, and S3E) recognizes a conserved peptide and was commercially obtained (Sigma, A2066 rabbit polyclonal). For the endogenous LIN-28 experiments (Figures 3A and S3E), we used an immuno-purified anti-LIN-28 rabbit polyclonal antibody that targets an epitope in the C. elegans LIN-28 C-terminus (Weaver et al., 2014). The anti-Flag epitope antibody (Figures 4B, 4E and S4C) was commercially obtained (Sigma, M2 mouse monoclonal). The anti-HA epitope antibody (Figures 4B–4F and S4D–S4E) was commercially obtained (Cell Signaling Technology, C29F4 rabbit monoclonal). The anti-V5 epitope antibody (Figures 4B–4F and S4C–S4E) was commercially obtained (Cell Signaling Technology, D3H8Q rabbit monoclonal).

For co-immunoprecipitation, cells were gently rinsed with 1× PBS then lysed (600 μl per well) with 10 mM Tris pH 7.4, 1 mM EDTA, 150 mM NaCl, 1% Triton-X-100, and protease inhibitors (Halt protease inhibitor cocktail, Thermoscientific) at 4°C for 30 minutes with gentle shaking. Lysates were spun at 16, 000 ×g for 20 minutes at 4°C. Supernatants were transferred to fresh tubes and total protein was determined by BCA assay (Pierce). Anti-HA conjugated agarose beads (Pierce) (Figures 4B–4F and S4C–S4E) or Anti-Flag M2 affinity gel beads (Sigma) (Figures 4B and S4C) were washed twice with TBS. Equal input of total protein was added to each immunoprecipitation and incubated with beads for 2 hours with rotation at 4°C. Beads were washed three times with TBST + 0.01% Tween-20. Proteins were eluted by addition of sample buffer and boiling for 4 minutes. Samples were resolved on a denaturing SDS-PAGE gradient gel, transferred by wet method, blocked and immunoblotted. Anti-Flag (Sigma M2, mouse monoclonal) or anti-HA (Cell Signaling C29F4, rabbit monoclonal) antibodies were used to detect CED-3 or its subunits and anti-V5 (Cell Signaling D3H8Q, rabbit monoclonal) was used to detect UBR-1. We found that tagged UBR-1 is resistant to direct immunoprecipitation by either Flag or V5 tags at the C-terminus or N-terminus. However, the tagged protein can be detected by denaturing conditions in Western blot either directly or following coIP with another interacting protein (Figures 4 and S4). We interpreted this as the result of steric hindrance due to buried residues in the folded structure.

In vitro N-end rule destabilization assay with chase and ATP regenerating system

The in vitro N-end rule assays shown in Figure 2 and S1D–S1E represent three experiments run independently. This assay was based on previous studies using the ubiquitin fusion method (Brower et al., 2013). Briefly, pWE517 (Asn 32 LIN-28) and pWE518 (Met 32 LIN-28) were used as templates to generate freshly synthesized L-³⁵S-Methionine-labeled substrates in vitro with a TNT kit (Promega) and used immediately. Destabilization assays were assembled following dilution of the TNT products such that 5 μl of TNT products were each diluted into 45 μl water, mixed well, and 5 μl of this 10-fold dilution was added to the 50 μl destabilization reaction containing the following components.

In a 20 μl volume, we assembled 25 mM Hepes pH 7.5, 5 mM MgCl₂, 0.25 mM DTT, 5 mM L-Methionine (unlabeled), 0.1 mg/mL cycloheximide, 12.5 mM ATP, 10 mM phosphocreatine, and 0.2 mg/mL creatine phosphokinase (rabbit muscle source, CPK first gently dissolved at 10 mg/mL in 100 mM imidazole, pH 6.6). The 20 μl mix including chase and ATP-regenerating system was added to 25 μl fresh rabbit reticulocyte lysate. Then 5 μl of the 10-fold diluted TNT products were added for a final reaction volume of 50 μl that was incubated at 37°C.

Aliquots of 5 μl were removed at 0, 15, 30, 60, 120 minutes and added to 45 μl of SDS loading buffer and mixed to stop the reaction. Samples were heated to 85°C for 5 minutes and 5 μl per well were resolved on a denaturing SDS-PAGE gradient gel, fixed for 30 minutes with 50% methanol and 10% glacial acetic acid. The gel was then incubated for 10 minutes in 10% glycerol then dried onto filter paper before phosphorscreen exposure. For quantitation, we used ImageJ-obtained values to calculate the percent of X-LIN-28 fragments remaining at the indicated times relative to the DHFR-Ub fragment internal control for the same time point.

In vitro CED-3 cleavage assay with purified enzyme

The CED-3 cleavage reaction was only used in this study to generate the N-end rule degradation test substrate shown in S2B. The CED-3 reaction was set up in the same manner as described previously (Xue et al., 1996; Weaver et al., 2014). LIN-28A was freshly synthesized with L-³⁵S-Methionine in vitro with a TNT kit (Promega) and used immediately. The cleavage reactions (without CED-3 (control) and with CED-3) were incubated at 30°C in a thermocycler with heated lid for 6 hr. CED-3 alone only modestly cleaves LIN-28A (Figure S1E, second lane +CED-3 and −ARS).

Next, equal amounts of the with CED-3 and without CED-3 products were used as inputs for the in vitro destabilization assay with chase (similar to in vitro N-end rule assay above), with or without an ATP regenerating system (similar to above) for a total of four reactions [±CED-3 incubated LIN-28 and ±ATP regenerating system (ARS)]. Samples were collected at the indicated times (similar to above) then resolved on a denaturing SDS-PAGE gradient gel, fixed for 30 minutes with 50% methanol and 10% glacial acetic acid. The gel was then incubated for 10 minutes in 10% glycerol then dried onto filter paper before phosphorscreen exposure.

Alignments and Structure annotation

Uniprot sequences (2015) were examined for protein domains using InterPro (Jones et al., 2014; Mitchell et al., 2015). These sequences were then aligned with Jalview (Waterhouse et al., 2009). PyMol (Molecular Graphics System, Version 1.8 Schrödinger, LLC) was used to annotate the deposited PDB file 3NY1 as indicated.

Seam cell assay

Images shown in Figure 2E were taken with a fluorescent microscope equipped with Differential Interference Contrast (DIC) optics (Zeiss Axioplan 2) and a Zeiss Axiocam MRm. All images were taken at the same magnification for the same amount of time. Seam cells were counted for only one side (left or right) of each animal (either the top or left side depending on the orientation of the animal in the viewing field). GFP signal was pseudo-colored.

Vulva development assay

Vulva development was scored with both a Leica MZ 16F dissecting scope and a Zeiss Axioplan 2 equipped with DIC optics. Animals were scored for vulva protrusion (Pvl) or fully penetrant vulva as previously defined (Sulston and Horvitz, 1977; Ambros and Horvitz, 1984; Moss et al., 1997).

Adult alae assay

Adult alae were scored using a microscope equipped with DIC optics (Zeiss Axioplan 2). Alae were scored as normal, low quality (thin and rough), or gapped (discontinuous) as previously defined (Weaver et al., 2014). Animals with both low quality and gapped alae were scored as gapped. Similar to seam cells, only one side was scored for each animal.

Cell culture transfection

HEK 293 cells were used only as a protein expression system. Cells were maintained in a humidified cabinet at 37°C with 5% CO₂. Cells were cultured with DMEM supplemented with 10% FBS, 4 mM L-Glutamine, 100 Units penicillin per mL, 100 μg streptomycin per mL, and 0.25 μg amphotericin B per mL. Cells were seeded in 6 well plates and transfected 24 hours later using TransIT®-LT1 Transfection Reagent (Mirus) according to manufacturer’s recommendations. Cells were harvested 48 hours after transfection for co-immunoprecipitation.

DIC optics and measuring hypodermal LIN-28::GFP expression

Synchronous mid-larval stage animals were examined for hypodermal LIN-28::GFP expression in the following blinded manner. Gravid adults were spot-bleached onto fresh NGM plates seeded with OP50. The next day, L3 stage animals were transferred to an agar pad on glass slides with white-light illumination on a dissecting scope. Using polarized light on a microscope equipped with DIC optics (Zeiss Axioplan 2) at 630× magnification, bright-field images were taken of the gonad for measurement of animal staging (Figure 3C). All animals of the appropriate stage, based on gonad length, were then illuminated for GFP fluorescence of the hypodermis (in the same field of view used to image the gonad) for exactly 5 seconds (representative images shown in Figure 3D; GFP signal was pseudo-colored) and all animal data collected were used for LIN-28::GFP intensity calculations such that no fluorescence data were excluded. The mean grey values for individual LIN-28::GFP fluorescent cells were determined using Fiji (Image J). All LIN-28::GFP positive hypodermal cells in the image were quantified. The sum of mean grey values for each animal is depicted in Figure 3E as a function of gonad length. The sum of mean grey values was used since this best reflects the dynamic range of LIN-28::GFP expression.

QUANTIFICATION AND STATISTICAL ANALYSIS

Statistics

Statistical tests, exact sample sizes, and p values were reported throughout the study. All biological replicates were independently generated, collected and processed. All data collected were used and displayed in main or supplemental figures. Regarding sample sizes, for molecular or biochemical assays, independent experiments were run in triplicate based on the rationale that obvious effects could be quantified from three replicates. For genetic tests, sample sizes showing mean and mode values close to the median value were ideal. This corresponds to a sample between 20 and 30 for unambiguous phenotypes. For incompletely penetrant phenotypes, sample sizes of 50 to 100 were assayed to quantify the biological variability.

For randomization, animals were scored in the order they were found without skipping. For blinding in the LIN-28::GFP fluorescence intensity studies, animals were first picked under white light illumination (without GFP) then scored for stage (measured by gonad length) and finally each animal that fit the proper stage criteria was then illuminated for GFP intensity in the order they were found without exception. Throughout this study, no data were excluded. Due to the large biological effect sizes measured with the randomly selected samples, blinding was not necessary for other assays and all data were used in calculations. Non-normal statistical tests that do not rely on assumptions of normal distributions were used throughout the study with the exception of t-tests used for gel quantitation. All statistical tests are clearly stated throughout.

Variability is displayed throughout the study in the form of dot-plots, standard deviation, or Box-Whisker plots as indicated with the exception of vulva penetrance and alae formation which are displayed more appropriately as summed percentages of the total since these represent categorical data. In cases where a genetic background resulted in high variability of a phenotype, this was concluded to represent a relevant aspect of the phenotype and was therefore retained during analyses.

Supplementary Material

1

Table S1, Related to the STAR Methods section. Oligos used in this study.

Highlights.

• LIN-28 inactivation by CED-3 caspase requires the Arg/N-end rule pathway

• CED-3 and Arg/N-end rule E3 ligase UBR-1 act on LIN-28 in a non-additive manner

• UBR-1 is required for CED-3 to efficiently recognize and cleave LIN-28

• CED-3 forms a complex with UBR-1 for efficient coupling of their activities